HAL Id: hal-01385558https://hal.archives-ouvertes.fr/hal-01385558
Submitted on 23 Nov 2020
HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.
Efficient reluctance network formulation for modelingdesign and optimization of linear hybrid motor
E. Manaa Barhoumi, Frédéric Wurtz, Christian Chillet, Boubker Bensalah,Olivier Chadebec
To cite this version:E. Manaa Barhoumi, Frédéric Wurtz, Christian Chillet, Boubker Bensalah, Olivier Chadebec. Effi-cient reluctance network formulation for modeling design and optimization of linear hybrid motor.IEEE Transactions on Magnetics, Institute of Electrical and Electronics Engineers, 2016, 52 (3),�10.1109/TMAG.2015.2496730�. �hal-01385558�
Fig. 1. Double-sided hybrid motor.
Fig. 2. Geometrical dimensions.
TABLE I
LHM DESIGN SPECIFICATIONS
Efficient Reluctance Network Formulation for ModelingDesign and Optimization of Linear Hybrid Motor
E. M. Barhoumi1,2,3, F. Wurtz3, Member, IEEE, C. Chillet3,B. Ben Salah1, and O. Chadebec3
1National Engineering School of Tunis, University of Tunis El Manar, Tunis 1002, Tunisia2College of Engineering, Majmaah University, Al Majmaah 11952, Saudi Arabia
3Univ. Grenoble Alpes, G2Elab, F-38000 Grenoble, FranceCNRS, G2Elab, F-38000 Grenoble, France
The aim of this paper is to propose an efficient and unique reluctance network (RN) model for all translator displacements ofa linear hybrid motor. This model has to consider several translator positions to analyze the linear motor performances. The proposedRN model takes into account the magnetic characteristic and the flux leakage. Consequently, the proposed model gives precise resultsof the electromagnetic characteristic. The RN model considers an interesting solution to deal quickly with an optimization task ofa large number of constraints and parameters. The developed model is coupled with sequential quadrating programming methodfor an optimization process of the linear machine. The obtained results confirm the efficiency of this method for modeling andoptimization of the linear machine.
Index Terms— Electromagnetic analysis, magnetic flux, magnetic forces, optimization, reluctance network (RN).
I. INTRODUCTION
T
known as flux tubes. Each flux tube is associated witha magnetic reluctance, which depends on circuit dimension andthe magnetic behavior. The originality of this model consists
HE LINEAR hybrid motors (LHMs) consist of anefficient solution for precise linear displacements appli-
cations [1]. Many research papers work on the design andthe optimization of these machines [2]. The finite elementmethod (FEM) is generally used when precise results are indis-pensable [3]. However, this method needs large computationtime, limiting the number of parameters that can be takeninto account. Moreover, the FEM cannot be coupled withrapid optimization algorithms [4]. The second method usedin the design of linear machines is based on the reluctancenetwork (RN) model [3]–[5]. In this paper, we are concernedwith the development of a new RN model for modeling andoptimization of linear machines. A 3-D view of the studiedmachine is shown in Fig. 1. The main dimensions of theproposed LHM are shown in Fig. 2.
The double-sided LHM mechanical dimensions are summa-rized in Table I.
In Section II, a unique RN model for all translator positionsis presented and discussed. The comparison of results with thatof the FEM is presented in Section III. The optimization ofthe machine is presented and discussed in Section IV. Theoptimization task consists in reducing the global mass of themachine. The electromagnetic characteristics of the optimizedstructure are also presented and verified using the FEM. Theconclusions and perspectives are given in Section V.
II. UNIQUE RELUCTANCE NETWORK MODEL
FOR ALL DISPLACEMENTS
The concept is based on the subdivision of the magneticcircuit of the machine into a sufficient number of elements
Fig. 3. Construction of the RN.
in its ability to calculate the electromagnetic characteristicsfor various positions from unaligned to aligned positions usinga simple run and a unique topology of model. An elementarystructure of the LHM with the respective reluctances is shownin Fig. 3.
A. Reluctances of the Air Gap
The air gap flux tubes are composed of straight lines witha constant length. However, the section of the air gap increaseswith the displacement x . Hence, the air gap reluctance isparameterized with the displacement, and it is given by thefollowing expression:
�e = e
μ0 · x · Tt. (1)
The salient in the mobile canalizes a non-neglected part of themagnetic flux. Hence, the salient is modeled by reluctancesexpressed as follows:
�s = Tpl + e
μ0 · (Tw − x) · Tt. (2)
B. Reluctances of the Stator
Generally, the magnetic reluctance of a ferromagneticmaterial, with a length l and a section S, can bewritten as
�fer = l
μ0 · μr · S(3)
μr is the relative permeability, which depends on the sat-uration of the ferromagnetic material. Hence, the ferromag-netic circuit of stator is divided into three different magneticreluctances connected in series. These reluctances dependon both the stator geometrical parameters and the magneticpermeability.
C. Reluctances of the Permanent Magnet
Each permanent magnet inserted in the stator is modeledby a magnetomotive force and a reluctance parametrized withthe geometrical and magnetic parameters of the permanent
magnet. This magnetomotive force and the reluctance of thepermanent magnet can be written, respectively, as
Ea = Br
μ0 · μaLa (4)
�a = 1
μ0 · μa
La
Sa(5)
where μa is the relative magnetic permeability of thepermanent magnet. La and Sa are, respectively, the length andthe section.
D. Reluctances of the Translator Part
The translator part is moved from unaligned to alignedpositions. Consequently, the translator canalizes different fluxtubes, which depend on the translator position. The RN modeldeveloped must be available for each translator position.Hence, the translator part is modeled by different reluctances.The flux in the second translator teeth, which must be alignedwith those of the stator in the aligned position, increases withthe displacement. Hence the reluctance of the teeth is inverselyproportional to the displacement, and it can be written as
�21 = Tpl
μ0 · μr · x · Tt. (6)
Applying the same methodology, the reluctances of other partsare determined. The reluctances �11 and �31 are, respectively,given as
�11 = Tpl
μ0 · μr · Tw · Tt(7)
�31 = Tpl
μ0 · μr · (Tw−x)2 · Tt
. (8)
The reluctances of the cylinder head depend on both the teethposition and the displacement. The reluctances �12, �13 and�22 are, respectively, given as
�12 = Tch
μ0 · μr · Tw · Tt(9)
�13 = Tch
μ0 · μr · (Tw−x)2 · Tt
(10)
�22 = Tch
μ0 · μr · x · Tt. (11)
The reluctances �23 and �32 are, respectively, given as
�23 = �32 = Tch
μ0 · μr · (Tw−x)2 · Tt
. (12)
E. Reluctances of the Leakage Flux
The RN model developed in this paper has two advantages.Firstly, both flux leakages between the stator and the translatoron one part and on the stator on the other are consideredand represented by the circumference arcs. This leakage fluxstrongly influences the flux with a considerable effect on themagnetic field in this actuator. In fact, RN simulation resultsobtained by neglecting the leakage flux lead to important errorscompared with the FEM. Secondly, the flux leakage reluc-tances vary with the translator displacement. The leakage flux
Fig. 4. Geometrical view of leakage flux tube.
Fig. 5. RN model and finite element model of 1/4 LHM.
is modeled by a magnetic reluctance given by the followingexpression:
�f = θ
μ0 · Tt
1
ln(
RexRin
) (13)
where θ is the angle of the circumferences arc. Rex and Rinare, respectively, the exterior radius and interior radius of thecircumferences arc, as shown in Fig. 4.
Each part of the flux leakage is modeled by an adaptedmagnetic reluctance. The parameters of the magnetic reluc-tances are parameterized with the displacement of the mobilepart. Indeed, the flux path of each part increases or decreaseswith the displacement of the mobile part. In this RN model,the leakage flux reluctances change according the translatordisplacement. The RN is modeled in the software Reluctool,and the finite element model of 1/4 LHM is shown inFig. 5.
When the translator is moved from the unaligned position tothe aligned position, the flux modeled using the reluctance �f 2decreased considerably. The angle formed is equal to 1.57 rad.The interior radius is constant and equal to the half of theair gap. However, the exterior radius decreases accordingthe expression Tw − x . The �f 2 is given by the followingexpression:
�f 2 = 1, 57
μ0 · Tt
1
ln(
Tw−xe/2
) . (14)
The exterior radius of the flux modeled using �f 4 decreasesaccording the expression (Tw − x)/2. Indeed, in the unalignedposition, the exterior radius is equal to the half of the toothwidth. Applying, the same methodology, all of flux leakagereluctances is determined. The �f 3 and �f 5 are, respectively,
Fig. 6. RN model algorithm.
Fig. 7. Propulsive forces calculated with FEM and RN model.
given as
�f 3 = 1, 57
μ0 · Tt
1
ln(
Twe/2+x
) (15)
�f 5 = 1, 57
μ0 · Tt
1
ln(
Tw/2e/2+x/2
). (16)
The �f 4 and �f 6 are written as
�f 4 = �f 6 = 1, 57
μ0 · Tt
1
ln(
(Tw−x)/2e/2
). (17)
The reluctance �f 1 represents the flux leakage canalizedbetween the two poles in the same stator. The flux leakagein this case is constant. The angle formed is 3, 14 rad.Consequently, �f 1 is given as
�f 1 = 3, 14
μ0 · Tt
1
ln(
1,5·Twe/2
). (18)
III. STATIC FORCES COMPARISON
The algorithm that solves the RN is shown in Fig. 6.It allows reducing the computation time. A set of mobile partpositions has to be defined before generating the model.
The propulsive forces calculated with RN model imple-mented by the Reluctool software and the FEM for threedifferent current excitations are shown in Fig. 7. As shownearlier, force curves are almost identical. Hence, the RN modelof the LHM is in accordance with the FEM.
IV. LHM OPTIMIZATION
The objective function of this optimization consists inminimizing the LHM mass while keeping the propulsive force.Moreover, the LHM mass is a very important parameterin the level of actuator manufacturing and transportation.However, many constraints must be respected. In particular,the propulsive force must be kept greater than 5000 N.The optimization constraints are summarized in Table II.
The RN model proposed in this paper reduces the compu-tation time in both modeling and optimization of the LHM.
TABLE II
OPTIMIZATION CONSTRAINTS
TABLE III
OPTIMIZATION RESULTS OF GEOMETRICAL PARAMETERS
Fig. 8. Initial and optimized LHM geometry.
The optimization process ends in 3463 ms after 28 iterations.The initial and optimal geometrical and electrical parametersare given in Table III.
The initial and optimized geometrical structures are shownin Fig. 8.
The mass of different parts of the initial and optimized struc-tures is summarized in Table IV. The maximum propulsiveforce is also given in the same table.
The propulsive force of the optimized structure is calculatedusing two methods: 1) the RN model and 2) the FEM.The obtained results are shown in Fig. 9, which confirms therobustness of the developed RN model.
TABLE IV
OPTIMIZATION RESULTS OF MASS AND PROPULSIVE FORCE
Fig. 9. Static force of the optimal structure calculated with FEM and RN.
V. CONCLUSION
A precise RN model is developed to model, design, andoptimize a linear actuator. This model is developed usingthe Reluctool software and is validated using the FEM.The efficiency of this model is its ability to generate automat-ically the electromagnetic characteristics for any mobile partpositions. This model has a good numerical stability and leadsto precise force computation. The model is applied for siz-ing by optimization. The sequential quadrating programmingmethod is used, and accurate optimization results are obtained.The simulation results presented and discussed confirm therobustness of the developed RN model and the optimizationapproach.
ACKNOWLEDGMENT
This work is supported by University of Tunis El Manar(Tunisia), Univ. Grenoble Alpes, G2Elab (France), and Alma-jmaah University (Saudi Arabia).
REFERENCES
[1] R. Krishnan, Switched Reluctance Motor Drives: Modeling, Simulation,Analysis, Design, and Applications. Boca Raton, FL, USA: CRC Press,2001.
[2] L. Szabó, I. A. Viorel, I. Chisu, and Z. Kovács, “A novel double salientpermanent magnet linear motor,” in Proc. Int. Conf. Power Electron.,Drives Motion (PCIM), 1999, pp. 285–290.
[3] B. du Peloux, L. Gerbaud, F. Wurtz, V. Leconte, and F. Dorschner,“Automatic generation of sizing static models based on reluctancenetworks for the optimization of electromagnetic devices,” IEEE Trans.Magn., vol. 42, no. 4, pp. 715–718, Apr. 2006.
[4] T. Raminosoa, I. Rasoanarivo, F. Meibody-Tabar, and F. M. Sargos,“Time-stepping simulation of synchronous reluctance motors using anonlinear reluctance network method,” IEEE Trans. Magn., vol. 44,no. 12, pp. 4618–4625, Dec. 2008.
[5] H. Nguyen-Xuan, H. Dogan, S. Perez, L. Gerbaud, L. Garbuio, andF. Wurtz, “Efficient reluctance network formulation for electricalmachine design using optimization,” IEEE Trans. Magn., vol. 50, no. 2,Feb. 2014, Art. ID 7021504.